1
\$\begingroup\$

schematic

simulate this circuit – Schematic created using CircuitLab

Today I saw an LED dimmer product. The driving circuit was like in the schematic. It was a cheap one but the circuit looks inefficient to me. When the BJT is not conducting, R1 pulls the gate high and the MOSFET starts conducting. That's ok. When the BJT is conducting (high signal on the GPIO,) the BJT shorts the gate to gnd. That's actually a commonly used circuit, but a large amount of current (12V/150 = 80mA) flows through the gnd on R1. Is it okay for a commercial product?

The PWM frequency is seen as 500Hz on scope. Obviously, when they increase the R1, rise time of the MOSFET gets very high and switching losses dramaticly increase. They claim that this product can handle up to 20 amperes (of course there is a heatsink on te MOSFET.)

Is there a efficient solution to drive a gate without a gate driver? Is this the right way to drive a MOSFET?

I don't have the product's document but here are the BJT and MOSFET datasheets.

\$\endgroup\$
2
  • \$\begingroup\$ In this circuit, with M1 switched on, it seems like there would be nothing to limit the current in the LED to a safe value except the current limit of the power supply or the Rds of M1. Maybe in the real circuit there is something else. But this further supports the idea that this circuit is sub-optimal (or was optimized to reduce cost above all else). \$\endgroup\$
    – user57037
    Commented Sep 26, 2019 at 23:36
  • 1
    \$\begingroup\$ in the actual circuit m1 is open drain. you can connect led strip to it. they have built in current limiting resistor on them. \$\endgroup\$
    – sinany1
    Commented Sep 27, 2019 at 6:32

3 Answers 3

4
\$\begingroup\$

For a penny or two more you can drive the MOSFET with a push-pull circuit.

schematic

simulate this circuit – Schematic created using CircuitLab

In this circuit Q1 and R1 level-shift (and invert) the input to drive Q2 and Q3 which are connected as complementary emitter-followers. Peak turn on current is limited by Q3 hFE and R1, so if R1 is (say) 4.7K, the current is in the hundreds of mA.

As far as the circuit you've got, if you have a relatively high load current, however, the 150 ohms may be considered justifiable. If the MOSFET is spending most of its time 'off', however, it's perhaps wasting significant power. Keep in mind the engineer who designed it was probably more interested in keeping their job than making the circuit a little bit better- the cost of those parts might pay for his or her salary.

The transistors are called upon to do two things- to drive the relatively huge gate charge of the MOSFET and perhaps to level-shift the input if it's less than Vdd. If the dimmer output was coming directly from a chip such as LM555 runing at 12V then there would be no need of any drive circuit.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ pretty clear. thanks! \$\endgroup\$
    – sinany1
    Commented Sep 26, 2019 at 21:35
1
\$\begingroup\$

I use the 1st circuit for years now for 24V PWM dimming of LED's. It works OK for 500Hz down to 1%, but when you go lower than 1% or increase the frequency you need the 3-transistor circuit. Note that the fist transistor (Q1) off-time will be the bottle neck. This is because the electrons in the silicon need to passively diffuse until it is off. Imagine it as a box which you spray with vapor, it takes some time to clear up the air. To speed this up we can add a ~220 picofarad speed-up base-capacitor parallel with R2, like some air pressure to blow the vapor away ;-)

\$\endgroup\$
0
\$\begingroup\$

Whoever designed that circuit made some newbie errors ignoring the series resistance and power dissipation of R1. — Let’s assume the LED is a strip with current limiting resistors suitable for 12V.

The input capacitance of M1 Ciss= 1470 pF and R1=130 ohms results in a power hungry current and unnecessary fast time constant =RC = 1.9us.

Considering Tau is 60% of Vcc and the range of Vgs from Ic= “off to on” is only about 15% here that further reduces the transition time.

There is a rule of thumb for cascading current buffers by current or impedance gains typ =1000 for FETs and 50 for BJT’s

Recommendation

These are commonly sold 5pc $8 online.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.